Research and development (R&D) is the creation of knowledge to be used in products or processes. Table 1 gives a summary overview of postwar U.S. R&D activity performed in industry. The first column gives privately financed R&D (PR&D) conducted in industry in billions of 1982 dollars. The second column gives the ratio of PR&D to investment in plant and equipment (P&E). The third column gives the share of federally financed R&D (GR&D) as a fraction of the total R&D in industry. State government and private nonprofit financing of basic scientific research that is part and parcel of teaching in colleges and university is not considered R&D. The only financing of research at universities and colleges that is considered R&D is R&D contracts to those institutions. Total university and college R&D in the sixties was 10 percent of the total R&D conducted in industry; in the eighties it was 13 percent.
Two facts stand out in table 1. First, investment that takes the form of R&D is growing relative to investment in P&E. Investment in P&E is recognized as investment by the official economic measurements; investment in R&D is not so recognized. Second, the role of government R&D is falling in relative terms.
There are several important issues in the economic analysis of R&D:
1. Is private R&D productive?
2. Is government R&D productive?
3. Is special government treatment for private producers of R&D justified?
4. Why is some government R&D so successful, while other government R&D fails?
5. Who benefits from U.S. R&D?
Is Private R&D Productive?
Beginning in the sixties, economists performed empirical tests confirming that investment in private R&D yields a positive return. This finding holds up for studies of R&D in general and in particular industries. Recent findings by Lichtenberg and Siegel reported an estimated rate of return of 35 percent for company-funded R&D. The older literature they surveyed reported an average rate of return of 29 percent. This is evidence of remarkable stability in the estimates of the rate of return to privately funded R&D. When Lichtenberg and Siegel decomposed R&D into basic and applied, they found that the rate of return to basic R&D was 134 percent, compared to the two older findings of 178 percent and 231 percent. When the rate of return, even after falling, is still in triple digits, one suspects underinvestment.
Is Government R&D Productive?
Econometric research almost never finds government R&D productive. Yet technical economists have long known about the remarkably high rate of return to agricultural GR&D. According to Robert Evenson, Paul Waggoner, and Vernon Ruttan, rates of return for government-financed agricultural R&D are consistently around 50 percent per annum. Ordinary people were able to see the efficacy of government-financed computers, electronics, and aviation in the Gulf War.
So why do broader studies find the opposite? One answer is as follows: Profit-maximizing companies use factors of production, whether they be labor, land, or R&D up to the point where their marginal value equals the marginal cost to the firm. But unlike wages paid to labor, the price that people pay to use government R&D is zero: one need only buy a technical journal to learn R&D results that cost millions to produce. Because companies pay zero for government R&D results, they use them up to the point where the marginal value equals zero. Economists looking for a positive marginal value of government R&D, therefore to find it. But all this means is that companies are using it a lot and that, while the marginal value of government R&D is zero, its total value is high.
There is a lively debate about whether government R&D enhances the supply of private R&D. The majority of economists, perhaps, hold that it does. Why would it? Because increasing the supply of one factor of production generally increases the marginal product of other factors. (More land, for example, makes a farm laborer more productive.) Similarly, more government R&D is likely to make private R&D more productive.
Is R&D Worthy of Special Treatment?
Knowledge epitomizes a public good. If someone produces knowledge, someone else can use it without paying for it. Therefore, the person who produced it will not be able to collect the full value of the knowledge produced. For this reason an unregulated, unsubsidized free market is likely to underproduce knowledge. As a result, most economists favor the creation of temporary monopolies through a patent system, such as the one provided for in the U.S. Constitution. With the prospect of a patent as a reward for innovation, people have more of an incentive to produce knowledge.
Need the government do more? Since some new knowledge is not patentable, perhaps special treatment is justified to encourage the provision of knowledge. The most dramatic case for special treatment is based on a famous argument made by Joseph Schumpeter. Schumpeter maintained that a monopoly—because it is able to garner more of the benefits to the industry from R&D (because a monopoly is the industry)—will have an incentive to invest more heavily in R&D than would a competitive industry. In economic jargon a monopoly can internalize more of the R&D benefits than a competitive industry can. Although Schumpeter himself did not argue for special treatment of R&D on this basis, the argument could be made. This consideration did not save the Bell system from breakup.
A much more modest argument—to give R&D tax credits—has been politically successful. But it is hard to tell whether the tax credit has been economically successful—that is, whether it has spurred private investment in R&D. One reason for not knowing the effect on R&D is that companies can get the tax credit simply by relabeling non-R&D expenditures as R&D. Nonetheless, the remarkably high rates of return to R&D that a wide range of studies report strongly suggests that there is underinvestment in R&D. Unfortunately, these studies do not allow one to suggest how to stimulate more R&D.
Why the Range of Government Experience?
If the experience with government R&D were uniformly wonderful or uniformly disastrous, students of R&D could offer easy guidance. However, the experience has been mixed. As mentioned, agricultural R&D and defense R&D in computers, electronics, and aviation have been remarkable successes. Balancing the accounts, one need only mention the supersonic transport, which was financed by British and French taxpayers, and the synfuels project, financed by U.S. taxpayers. The costs for each of these projects exceeded the benefits by billions of dollars. Yet making a list of winners and losers is somewhat beside the point when one of the winners, the computer, has changed the world.
This list of failures raises a question: if the government can pick winners in defense, why not elsewhere? It is important to note that in aircraft and electronic R&D, the Defense Department was the major customer for many years. This is in the context of a political decision not to match the buildup of the late Warsaw Pact man for man and tank for tank. Rather, the Defense Department was charged with matching the Warsaw Pact with higher-quality equipment. The competing branches of the U.S. armed services could be held politically accountable for their performance. The resulting incentives seem to have made the Defense Department very sensitive to how infant technologies could be developed to serve its clearly delineated mandate. Similarly, agricultural R&D has long enjoyed a politically symbiotic relationship with agricultural interest groups. When government agencies have incentives to be competent, they are competent.
But who monitors R&D done only for the "public good"? The usual answer is no one. Simple public choice theory suggests that government responds to incentives. When the performance of government agencies is monitored carefully, one expects very different results than when no one in particular is supposed to benefit from the R&D expenditures. Thus, there is no reason to believe that the success rate of defense and agricultural R&D could be replicated in other areas.
Who Benefits from U.S. R&D?
One difference between stocks of knowledge and stocks of physical capital is that stocks of knowledge can be shared. If I build a machine, it cannot produce for you unless it stops producing for me. If I learn something, this knowledge can produce for you and for me at the same time. If this is so, then the rest of the world should be a major beneficiary of U.S. R&D. Other countries can rent the knowledge, or even get it for free, without having to create it themselves. This suggests that a program of high-tech economic nationalism is automatically self-defeating. One can, with difficulty, block the export of a machine. But the export of knowledge is much harder to impede.
David M. Levy is an economics professor at George Mason University.
Evenson, Robert E., Paul E. Waggoner, and Vernon W. Ruttan. "Economic Benefits from Research: An Example from Agriculture." Science 205 (1979): 1101-7.
Flamm, Kenneth. Targeting the Computer. 1987.
Levy, David M. "Estimating the Impact of Government R&D." Economics Letters 32 (1990): 169-72.
Levy, David M. "Public Capital and International Labor Productivity." Economics Letters 39 (1992): 365-68.
Lichtenberg, Frank R., and Donald Siegel. "The Impact of R&D Investment in Productivity." Economic Inquiry 29 (1991): 203-29.
Nelson, Richard R., ed. Government and Technical Progress. 1982.